摘要
特征选择是示例学习的关键 ,直接关系到获取的概念的优劣。基于扩张矩阵理论和粗集理论 ,将特征子集的选择问题转化为数学优化问题 ,提出了相应的优化模型。这种优化模型易于理解 ,采用现有的软件即可求解 ,克服了以前许多特征选择算法的不足。
Learning from examples is widely studied in machine learning because it is a very effective cure for the bottleneck problem of knowledge acquisition. To discern positive and negative example fully, feature subset selection plays a great role in learning from examples. The smaller the base of feature subset, the better it is for concept extraction, but the optimal feature selection has been proved to be a NP hard problem. There are many disadvantages in previous algorithms. Based on extension theory, which used to be utilized for heuristic algorithms, and rough set, which is especially suitable for reduct of decision tables, we change the feature selection into an optimization problem and the corresponding models are proposed. The models are both solved by existing software or genetic algorithms (GAs) and more understandable. The method above are used for a method for concept extraction in KDD (Knowledge Discovery in Database) and the result is satisfactory in addition to overcoming some disadvantages of previous algorithms.
出处
《系统工程理论方法应用》
2002年第2期153-156,162,共5页
Systems Engineering Theory·Methodology·Applications
